[
https://issues.apache.org/jira/browse/HIVE-14515?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15433749#comment-15433749
]
Prasanth Jayachandran commented on HIVE-14515:
----------------------------------------------
[~mmccline] I ran all the schema evolution tests without removing INSERT INTO
.. VALUES on MiniLlap with LLAP IO disabled. When LLAP IO is disabled, the
source input format (HiveIF) will be used by default. So no code path related
to LLAP reader will be used for schema evolution tests. With that all the 26
schema evolutions tests ran in 16mins. I think it is better to run them in
MiniLlap with llap io disabled so that regular code path is used. I think the
main benefit we are getting with llap is the cache + multiple executors. I
would recommend to move over the tests so that we don't have to fix this. What
do you think?
> Schema evolution uses slow INSERT INTO .. VALUES
> ------------------------------------------------
>
> Key: HIVE-14515
> URL: https://issues.apache.org/jira/browse/HIVE-14515
> Project: Hive
> Issue Type: Sub-task
> Components: Hive
> Reporter: Matt McCline
> Assignee: Matt McCline
> Priority: Critical
>
> Use LOAD DATA LOCAL INPATH and INSERT INTO TABLE ... SELECT * FROM instead.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
